Check please! AI expansion in radiology hinges on healthcare’s value proposition
Regulatory approval is the first hurdle for artificial intelligence adoption in the U.S., but payment represents a much bigger challenge. And overcoming the latter may prove difficult until healthcare matures into a value-based system, radiologists argued in a special report published Wednesday.
As of early 2020, about 72% of all FDA-approved AI in healthcare was related to radiology, researchers explained in Radiology: Artificial Intelligence. The U.S. Centers for Medicare & Medicaid Services has recently started reimbursing providers for two AI tools, but these payment pathways remain limited and may not entice private payers to follow the government’s lead.
“Sustained adoption of AI through the current reimbursement framework may be challenging in a fee-for-service environment,” Melissa M. Chen, MD, a radiologist at MD Anderson Cancer Center in Houston, and colleagues added. “As value-based payment models mature, in which measuring improvement in quality becomes increasingly important at decreased costs, AI becomes a valuable tool for radiologists and healthcare systems.”
Where radiology AI reimbursement currently stands
Just last year, CMS began paying providers for using two AI tools: ContaCT and IDx-DR. The former, created by Viz.ai, is paid for under the New Technology Add-on Payment pathway and alerts neurosurgeons when a CT scan shows evidence of a blood clot in the brain.
Digital Diagnostics’ IDx-DR tool, meanwhile, analyzes retinal images to diagnose diabetic retinopathy and is reimbursed via a CPT code in the Physician Fee Schedule. It is the first AI CPT code created by the American Medical Association CPT Editorial Panel.
Both pathways, however, are inherently limited and unable to capture the true value of AI, the authors explained.
For example, payment under NTAP is only available for 3 years and actual reimbursement is often misaligned with CMS’ projections. Technologies are also required to be different from existing tools, which may hinder new development, the group explained.
Similarly, some estimates predict CMS payment for IDx-DR wouldn’t cover the cost of the technology. Chen et al. further noted this AI code only applies when no physician work is done, discounting algorithms that augment tasks and those that require nuanced time changes in certain clinical situations.
The payment path forward
Chen and colleagues suggest AI reimbursement may begin to evolve as providers continue to participate in the 2015 Medicare CHIP and Reauthorization Act or MACRA, and its two value-based programs: the Merit Incentive Payment System (MIPS) and Alternative Payment Models (APMs).
Under MIPS—which scores physicians on 4 categories, including quality—AI could automatically extract quality metrics in an imaging report to bolster radiologists’ performance and increase bonus payments. Rads would be willing to pay for AI in this situation and could ultimately help develop meaningful metrics linked to improved outcomes.
The group also noted APMs may be the future for radiologists, as they focus on population health while shifting financial risk to physicians. Such algorithms could analyze imaging to forecast patients at-risk of diseases and help providers intervene earlier. Health systems or payers would likely embrace such tools to reduce costs and improve outcomes, the authors noted.
The radiology AI market is growing quickly and is projected to hit $2 billion by next year. For Chen and co-authors, the riddle of who will pick up the AI tab is complicated, but may ultimately come down to self-interest.
“The entity that receives the most benefit likely will pay for AI, and ultimately may consider this payment simply the cost of doing business,” the authors concluded.
Matt joined Chicago’s TriMed team in 2018 covering all areas of health imaging after two years reporting on the hospital field. He holds a bachelor’s in English from UIC, and enjoys a good cup of coffee and an interesting documentary.